A student can calculate 347 + 286 correctly.
Ask why a ten is regrouped, however, and the explanation becomes:
“Because that is what you do.”
Another student can explain perfectly that ten ones are equivalent to one ten, but takes so long to complete every addition that a multi-step problem collapses under its own arithmetic load.
Both students know something important.
Neither yet has the whole capability.
Mathematical proficiency needs meaning and movement: understanding the relationships, and being able to use them accurately, efficiently and flexibly.
This is why the old argument—understanding or practice? concepts or procedures?—is badly framed.
Strong mathematics needs both.
The quick answer: concepts explain the system; procedures let the learner operate inside it
Conceptual understanding means knowing the mathematical relationships well enough to explain, represent, connect and reason about them.
Procedural fluency means carrying out mathematical procedures accurately, efficiently and with enough flexibility to choose or adapt a method when appropriate.
They are different.
They are also deeply connected.
| Conceptual understanding asks | Procedural fluency asks |
|---|---|
| Why does this work? | Can I execute it reliably? |
| What relationship is being preserved? | Can I choose an efficient method? |
| How does this connect to another representation? | Can I perform the steps without losing signs, units or place value? |
| What would change if the problem changed? | Can I adapt the procedure when the surface changes? |
The learner needs the two columns to meet.
Example 1: regrouping is not “carry the one”
Consider:
47 + 28.
A procedural description might say:
- 7 + 8 = 15;
- write 5;
- carry 1;
- 4 + 2 + 1 = 7;
- answer 75.
The answer is correct.
But the conceptual explanation is:
15 ones can be renamed as 1 ten and 5 ones.
So the extra “1” is not an arbitrary mark floating above the tens column. It represents one additional ten.
When that meaning is understood, the written algorithm becomes more stable because the learner knows what each recorded digit represents.
A procedure becomes safer when every step can be attached to a quantity or relationship.
But understanding without fluency creates its own bottleneck
Imagine a Primary 6 learner who understands ratio beautifully but still has to reconstruct basic multiplication facts slowly.
The ratio problem may require:
- finding total units;
- dividing a total by those units;
- multiplying to recover several quantities;
- then applying a second relationship.
If every multiplication or division consumes substantial attention, less working memory remains for the structure of the problem.
Fluency therefore matters not because speed is the purpose of mathematics, but because dependable lower-level procedures can release attention for higher-level reasoning.
Example 2: adding fractions exposes the difference
Consider:
1/3 + 1/4.
A learner may memorise:
“Find the LCM, change the denominators, add the numerators.”
That can produce:
4/12 + 3/12 = 7/12.
Conceptually, the essential idea is that thirds and quarters are different-sized units. They cannot be counted together directly.
Equivalent fractions rename both quantities using a common unit: twelfths.
Once the units match, the quantities can be combined.
This explanation protects against the familiar error:
1/3 + 1/4 = 2/7.
The child who understands the unit structure can see why adding denominators creates a new unit that was never present in the problem.
Procedural fluency is not the same as rushing
Fast work can be fluent.
Fast work can also be brittle.
A learner who races through one familiar format but fails when the numbers, representation or unknown change has not developed robust fluency.
Useful fluency includes:
- accuracy;
- reasonable efficiency;
- appropriate method selection;
- flexibility among valid methods;
- ability to monitor whether the answer makes sense.
Speed is one possible consequence, not the whole definition.
Example 3: long division needs both structure and routine
A written long-division algorithm compresses repeated reasoning about place value, grouping and remainders.
Without conceptual understanding, a learner may:
- place a quotient digit in the wrong column;
- forget what the remainder represents;
- continue a decimal without understanding the place-value extension;
- accept an answer larger than the dividend when the situation makes that impossible.
Without procedural fluency, however, the learner may understand division as grouping but become lost inside a four-digit calculation.
Good teaching therefore makes the place-value logic visible and then gives enough well-designed practice for the routine to become dependable.
Example 4: percentage reveals whether the learner owns the base quantity
A learner may be very fluent at finding 20% by multiplying by 0.2.
But consider:
A price increases by 20%, then decreases by 20%.
Does it return to the original price?
No.
If the original price is $100:
- after a 20% increase: $120;
- 20% of $120 = $24;
- after the decrease: $96.
The procedure is easy.
The conceptual issue is that the second percentage uses a different base.
This is a useful transfer test because a learner who merely associates “+20%, −20%” with cancellation may not own the percentage relationship.
Example 5: equation solving shows how concepts become procedures
Consider:
x + 7 = 19.
A weak procedural story says:
“Move the 7 across and change the sign.”
A stronger conceptual story says:
The equation states that both sides have equal value. Subtracting 7 from both sides preserves that equality.
x + 7 − 7 = 19 − 7.
x = 12.
With experience, a learner may compress the written steps.
But the compression is safe because it rests on a preserved invariant: equality.
Expert-looking shortcuts are safest when they are compressed versions of understood reasoning rather than unexplained rituals.
Concepts and procedures can strengthen each other in both directions
The relationship is not simply:
understand first → practise later.
Sometimes carefully designed practice reveals structure.
Comparing:
- 48 + 19;
- 48 + 20 − 1;
- 50 + 17
can deepen understanding of compensation while also improving mental fluency.
Similarly, practising several equivalent equations can make equality more familiar while conceptual discussion explains why the transformations are valid.
Learning can therefore cycle:
meaning → method → practice → pattern noticed → deeper meaning → more flexible method.
Representation is one bridge between the two
Concrete objects, diagrams, bar models, number lines, tables and equations can expose why a procedure works.
For example, a 4×6 array can show multiplication as equal groups and area while also supporting the fluent fact 4×6=24.
A fraction strip can show why 2/3 equals 4/6 while symbolic work makes equivalent fractions efficient to generate.
The goal is not to keep every learner permanently attached to a diagram.
The goal is to let the meaning survive when the representation becomes more compressed.
A four-part diagnostic separates different kinds of “I can do it”
When a learner appears to know a topic, test four things.
| Diagnostic | Question |
|---|---|
| Explain | Why does the method work? |
| Represent | Can the same relationship be shown another way? |
| Execute | Can the procedure be carried out accurately and efficiently? |
| Transfer | Can the learner recognise and use the idea when the surface changes? |
A student can succeed in one row and fail another.
That difference tells the teacher what to repair.
Diagnostic pattern 1: correct procedure, weak explanation
The learner gets ten fraction questions correct but cannot explain why common denominators are needed.
Do not remove all practice.
Add comparison tasks, visual models and error analysis that force the unit structure into view.
Diagnostic pattern 2: strong explanation, weak execution
The learner explains regrouping correctly but makes frequent arithmetic or transcription mistakes.
Do not reteach the concept from zero unless evidence shows it is unstable.
Use short, focused practice that builds accuracy and automaticity while preserving meaning.
Diagnostic pattern 3: familiar success, transfer failure
The learner can execute a method when the worksheet heading says “Percentage of a Quantity” but fails when the same relationship appears inside a multi-step word problem.
The procedure exists.
Recognition and representation are weak.
Use mixed practice and changed surface stories so the learner must identify the relationship rather than retrieve a chapter routine.
Diagnostic pattern 4: slow but thoughtful work
The learner reaches correct answers and explains them well, but every simple calculation is laborious.
Here procedural fluency is the leverage point.
Use spaced retrieval, carefully chosen fact practice and efficient strategy comparison without turning every lesson into a speed test.
Practice should not be designed only to repeat answers
Two sets of ten questions can produce very different learning.
Set A changes only the numbers.
Set B changes what matters:
- one problem asks for the whole;
- one asks for a part;
- one changes the representation;
- one includes irrelevant information;
- one presents a common misconception;
- one asks for two correct methods;
- one asks which method is more efficient and why.
Set B trains the procedure while also strengthening selection, explanation and flexibility.
Worked example comparison: two correct methods
Calculate 398 + 257.
Method 1: standard vertical addition.
Method 2: compensation.
398 + 257 = 400 + 255 = 655.
Both are correct.
The useful learning question is not merely which method is “best”.
Ask:
- Why does compensation preserve the sum?
- Which method is less error-prone for this learner?
- Would the choice change for 483 + 679?
- Can the learner predict when compensation will be efficient?
This turns fluency into strategic fluency.
Common misconception 1: conceptual teaching means avoiding algorithms
Algorithms are valuable mathematical tools.
The problem is not that they are procedures. The problem is when the learner cannot connect the procedure to mathematical meaning or use it appropriately.
Repair: explain the invariant, model the procedure, practise it, then vary the context.
Common misconception 2: fluency means memorising without understanding
Useful fluency includes flexibility, selection and monitoring, not merely rapid recall.
Common misconception 3: if a child can explain it, practice is unnecessary
Explanation does not automatically produce dependable execution under time pressure or inside multi-step problems.
Repair: add sufficient retrieval and varied application for the method to become stable.
Common misconception 4: many correct worksheet answers prove conceptual mastery
Repeated familiar formats can be solved by pattern matching.
Repair: change the representation, move the unknown, ask for explanation, or present a counterexample.
Common misconception 5: every learner needs the same balance at the same moment
One learner may need the concept reopened.
Another may need short fluency practice.
A third may need transfer tasks.
The correct balance is diagnostic rather than ideological.
A diagnostic ladder for concepts and fluency
- Can the learner calculate a familiar example?
- Can the learner explain why the procedure is valid?
- Can the learner show the relationship with a second representation?
- Can the learner detect an incorrect worked example?
- Can the learner choose between two valid methods?
- Can the learner execute accurately without excessive cognitive load?
- Can the learner retain the skill after a delay?
- Can the learner use it inside a mixed-topic problem?
- Can the learner adapt when the unknown or representation changes?
- Can the learner check whether the result is reasonable?
How this fits Singapore Primary Mathematics
The current MOE Primary Mathematics syllabus places mathematical problem solving at the centre and combines content with processes such as reasoning, communication, connections, applications and modelling. That architecture is consistent with developing knowledge that is both understood and usable rather than treating procedures and reasoning as separate subjects.
The syllabus also emphasises foundational numerical skills and the ability to apply concepts in varied contexts. In practice, this means learners need reliable procedures and the conceptual relationships that let those procedures transfer.
Research and practice guidance outside Singapore makes a similar distinction. The Institute of Education Sciences separates conceptual knowledge, procedural knowledge and procedural flexibility when evaluating mathematical problem solving, while current Education Endowment Foundation guidance on manipulatives emphasises using representations to reveal structure while supporting flexibility and fluency.
What parents should ask after homework
Instead of choosing between:
“Did you understand it?”
and:
“How many questions did you finish?”
ask for evidence from both sides:
- “Show me why that step works.”
- “Can you do a second example without help?”
- “Can you show it another way?”
- “Which method would you choose if the numbers changed?”
- “How would you notice if your answer were impossible?”
The answers reveal whether the learner owns only the explanation, only the routine, or an increasingly integrated mathematical capability.
The deeper lesson: fluency is compressed understanding at its best
When mathematics is first learned, many relationships need to be made explicit.
Why ten ones become one ten.
Why fractions need common units before addition.
Why equality survives when the same operation is applied to both sides.
With successful practice, some of that reasoning becomes compressed into efficient procedures.
The compression is useful as long as the meaning remains recoverable when a new problem demands it.
The strongest learner is not the one who chooses understanding over fluency, but the one whose fluency is grounded in relationships and whose understanding can still operate efficiently when the mathematics becomes demanding.
Connected eduKateSG Mathematics routes
- Translating Between Words, Models, Diagrams, Tables and Equations
- Primary 4 Mathematics Diagnostic
- Primary 5 Mathematics Diagnostic
- Primary 6 Mixed-Topic Transfer Diagnostic
- Regrouping in Addition: What Carrying Really Means
Sources and further reading
- Singapore Ministry of Education — Primary Mathematics Syllabus, updated October 2025
- Institute of Education Sciences — Improving Mathematical Problem Solving in Grades 4 Through 8
- Institute of Education Sciences — Arithmetic Practice that Promotes Conceptual Understanding and Computational Fluency
- Education Endowment Foundation — Using Manipulatives to Develop Mathematical Understanding